#: 20925 S1/General Interest 25-Apr-95 14:39:15 Sb: OS/9 bulk disk copy? Fm: Hans Van Schepdael 100530,1015 To: all Dear, My disk duplication facility has a problem! I want to duplicate OS/9 floppies on a TRACE copier. The copier, however, can't read those disks. Are there drivers available for the TRACE copiers so that OS/9 disks can be duplicated. Many thanks for the help. Hans #: 20934 S1/General Interest 27-Apr-95 12:30:30 Sb: DAVID System Fm: Ugo Alvazzi del Fra 73364,3312 To: all Does anybody know something about DAVID system ? What I know actually is: - runs OS/9 - supports the RAVE environment I would appreciate much any further information. Thank you, Ugo Alvazzi del Frate Rome ITALY #: 20935 S1/General Interest 28-Apr-95 14:04:23 Sb: CoCo Fest Update Pt#1 Fm: Rogelio Perea 72056,1204 To: ALL NEWS RELEASE NEWS RELEASE NEWS RELEASE Here are the 5 "W's" WHO? 1) The Glenside Color Computer Club of Illinois presents WHAT? 2) The Fourth Annual "Last" Chicago CoCoFEST! WHEN? 3) April 29th & 30th, 1995 (Sat. 10am-6pm; Sun. 10am-4pm) WHERE? 4) HOLIDAY INN ELGIN (A Holidome Indoor Recreation Center) 345 W. River Road (A city block from I-90 & IL-31S) Elgin, Illinois (Same great location as last year!) Overnight room rate: $55.00 (plus 10% tax) Call 1-708-695-5000 for reservations. Be sure to ask for the "Glenside" or "CoCoFEST!" rate. >>>>> YOU MUST REGISTER UNDER COCOFEST TO GET THIS RATE <<<<< WHY? 5) A. To provide vendor support to the CoCo Community B. To provide Community support to the CoCo Vendors C. To provide educational support to new users. D. TO HAVE AN OUTRAGEOUSLY GOOD TIME!!!!! And now, the "H" word. HOW MUCH? 1) Admission: $8.00 at the door, 2-day pass $5.00 at the door, 1-day pass Advance ticket sales: $5.00 + SASE for the whole show or $5.00 + $.50 postage & handling Children 8 and under FREE Contact: George Schneeweiss, Treasurer Glenside Color Computer Club RR#2 Box 67 Forrest, IL 61741-9629 For further information, general or exhibitor, contact: Tony Podraza, Fest Chairman Rob Gibons, Vice President, GCCCI 708-428-3576, VOICE 708-916-1118, VOICE 708-428-0436, BBS Eddie Kuns, President, GCCCI 708-820-3943, VOICE #: 20936 S1/General Interest 28-Apr-95 14:04:40 Sb: CoCo Fest update Pt#2 Fm: Rogelio Perea 72056,1204 To: ALL ******************************************************************** * * *************************************************************** * * * * * * * FEST UPDATE !!!! FEST UPDATE !!!! FEST UPDATE !!!!! * * * * * * * * MARTY GOODMAN WILL BE AT THE FEST GIVING SEMINARS * * * * * * * * ALSO THERE WILL BE: * * * * Brother Jeremy Mike Cary * * * * No Minimum Bid Auction on both days * * * * Boise CDI KD Consulting * * * * OS9 Users Group Meeting Nitros OS9 * * * * Open OS9 forum Glenside Meeting * * * * And Much Much More..... * * * * * * * * ALSO IF YOU'RE LOOKING TO GET A ROOM FOR THE FEST, * * * * ALL ROOMS AT THE HOLIDAY IN ARE TAKEN; * * * * HERE ARE THE OTHER HOTELS IN THE AREA: * * * * SUPER 8 ELGIN INN BEST WESTERN * * * * 708-697-8828 708-695-3000 708- - * * * * HAMPTON INN AND DOWN THE ROAD A HOWARD JOHNSON * * * * 708-931-1940 708-695-2100 * * * * * * * *************************************************************** * ******************************************************************** * #: 20940 S1/General Interest 02-May-95 03:22:35 Sb: MVME162 Fm: Roy D Miller 100354,3307 To: ALL Help I have a MVME162 board and I am having problems initialising and using the Z85230 serial controlers. Has any one any source code that would point me in the right direction. Thank if you can help Roy Miller #: 20942 S1/General Interest 05-May-95 10:11:11 Sb: CD-I RTF GENERATOR Fm: CHRISTOPH STUMMER 100067,1570 To: ALL Who knows a Software named RTF GENERATOR ???? What is this Software doing ???? Which firm distribute this Software ??? Thanks Stummer #: 20972 S1/General Interest 30-May-95 10:43:39 Sb: X.25 & LAP-B Fm: Paul S. Dayan 100040,400 To: All, Does anyone know of X.25 and/or LAP-B support software for OS-9 V3.0 ? #: 20969 S5/OS9 Users Group 24-May-95 15:19:43 Sb: #20763-In search or a COCO III Fm: Don A. Dorsey 73204,3140 To: - Visitor 75262,1257 I have a CoCo 3 that a may be willing to part with. email me Don Dorsey 73204,3140 #: 20951 S6/Applications 12-May-95 15:55:00 Sb: #Ved Upgrades Fm: Rick LaVallie 76653,2244 To: Bob Van der Poel/ 76510,2203 (X) I would be interested in buying any upgrades you have done to Ved and Vprint I bought the package several years ago at an 0s9 meeting at my house you probably do not remember being there,Bruce Isted and Jerry Mclary were both preasent at the time. Any way I've just installed your HDISK.AR and it seems to be working better/faster than the old one are there any other improvenments to the SCII & 4 in 1 Mods. around? - please get back to me- Rick LaVallie There is 1 Reply. #: 20954 S6/Applications 12-May-95 22:45:07 Sb: #20951-#Ved Upgrades Fm: Bob van der Poel 76510,2203 To: Rick LaVallie 76653,2244 (X) Yes, Rick, I _do_ remember the meeting. Was fun. The latest edition numbers are: ved 2.9 vpt 2.1.6 upgrades are available to registered users for $5.00 each. Just send me or original disks in a mailer with a check and a couple of bucks for postage and I'll fire them back to you. Nice to hear from you again. There is 1 Reply. #: 20957 S6/Applications 15-May-95 07:48:20 Sb: #20954-#Ved Upgrades Fm: Rick LaVallie 76653,2244 To: Bob van der Poel 76510,2203 (X) Hello Bob nice to here form you . I don't seem to have your mailing address anywere or at least i cannot find it could you send it to me thanks. On another subject I'm sure Bruce would like me to say Hi for him not sure if you have talked to him lately but he is on his way to the Sudan today at 18:00,should be an interesting trip. Will send my origanal VED disk to you as soon as I get your address Thanks Rick LaVallie. There is 1 Reply. #: 20960 S6/Applications 15-May-95 22:25:33 Sb: #20957-#Ved Upgrades Fm: Bob van der Poel 76510,2203 To: Rick LaVallie 76653,2244 (X) From Canada, please send mail to: PO Box 57 Wynndel, BC V0B 2N0 No, I haven't talked to Bruce for quite some time. I was going to call him the last time I was in Calgary...but never got to it. No Time, No Time! So, is he going to do all kinds of OSK integration in the desert? There is 1 Reply. #: 20961 S6/Applications 17-May-95 10:02:08 Sb: #20960-#Ved Upgrades Fm: Rick LaVallie 76653,2244 To: Bob van der Poel 76510,2203 (X) Thanks for the address Bob I will send this out today. Do you know where I might find an upgraded version of OSTERM I'm useing version OSTERM.142 and was hopeing there was a newer version around or even a better terminal program for the COCO. As for Bruce I do not think the desert is much of a desert this time of year more than likely he'll be playing in the mud as i believe he said it was the rainy season over there. Rick LaVallie There is 1 Reply. #: 20966 S6/Applications 21-May-95 14:07:21 Sb: #20961-Ved Upgrades Fm: Bob van der Poel 76510,2203 To: Rick LaVallie 76653,2244 >Do you know where I might find an upgraded version of OSTERM I'm useing >version >OSTERM.142 and was hopeing there was a newer version around or even a better >terminal program for the COCO. You might want to ask Colin McKay. You can reach him at cmckay@northx.isis.org. #: 20963 S6/Applications 18-May-95 00:16:57 Sb: #Help needed Fm: Ken Drexler 75126,3427 To: All I have written an editor and word processor for OSK and am looking for several people to try it out. The editor uses a terminal for input and, within the limits of the terminal, is a what-you-see-is-what-you-get type editor. It can edit documents where each line is independent of each other, such as computer code as well as documents where the lines blend together into sentences and paragraphs. I have used the editor for over a year for writing programs (including the editor) and letters and other documents on my Kix\30 using Wyse 150 terminals. I would like to learn whether the program runs as well on other OSK machines and terminals. (I do not know, for instance, how it runs on a MM1 machine or with Kimtron or other terminals.) Also, writing an editor involves choices about the user interface and the commands and single-key operators which are available in it. I need feedback on whether the current features are useful, whether they work as expected and whether any needed features have been omitted. If you have an OSK machine and are willing to take a test drive and tell me what you think, send me email saying so. Ken Drexler 75126,3427 There is 1 Reply. #: 20965 S6/Applications 18-May-95 23:02:57 Sb: #20963-Help needed Fm: SCOTT HOWELL 70270,641 To: Ken Drexler 75126,3427 (X) I have a CDS020 68020 system and will be glad to test the software on my system #: 20967 S6/Applications 21-May-95 15:05:55 Sb: Zip Bug fixed(?) Fm: David Breeding 72330,2051 To: all Hey, everyone. I guess everyone remembers me crying a while back about ZIP messing up on the directory structure. I kept getting occasional instances when the bitmap for the file was not allocated. The version I have is 1.0 Someone said it was buggy, but I have had pretty good luck in unzipping (different program, of course), other files and have had no complaints on files I've zipped. I even grabbed a new zip (1.9) from chestnut, but _it_ seemed to do mess up more often than 1.0. I finally went in and began studying the code that came with zip, and I _believe_ I have discovered the problem.. When all files are zipped, it has created a new temp file, usually with the name _Z00*. On exit, it first deletes the original *.zip, if present. Next, it goes to a function called link(). It attempts to overwrite the temp filename with that of the destination *.zip name, inc the link count in the file descriptor. (Otherwise, if any part fails, it comes back and does a straight copy). In either case, it calls unlink() to delete the tempfile name. I found that link() failed often, and the moderately rare cases when link() was successful, were when the file structure got munged. The problem: I believe I've found it. In the file "osk.c", the main function that link() uses is a multi-purpose one, and it optionally will get the sector size from the device driver, or will read the disk. Two defines are used to control the program flow. "FROM_DRIVER" and "FROM_DISK". My code defines FROM_DISK, but... somehow, the author, down around line 750, when he wants to read the disk, uses the switch #ifdef "FROM_MEDIA" instead of "FROM_DISK". When I changed this to FROM_DISK, it appears in my test program that link() is working much more often (most of the time(?)) and that the few times I have tried it, it is now fixed... (I was a little chicken to try it on my HD - but did) If anyone is using zip, it might be well to fix this. Actually, I wonder if it would be safest to just eliminate this link() call from the procedure, as I wouldn't mind its doing a copy, as my main use for it is in small QWK replies and large copies to floppy for MS-DOS systems is not all that frequent. Is there someone who should be notified about this? I couldn't determine who might be in charge of the OS-9 end of it.. -- David Breeding -- CompuServe : 72330,2051 Delphi : DBREEDING *** Composed with InfoXpress/OSK Vr. 1.02 & VED Vr. 2.4.0 *** #: 20937 S8/BBS Systems/TSMon 28-Apr-95 14:05:05 Sb: BBS Info wanted! Fm: Rogelio Perea 72056,1204 To: ALL Here's a cross-post from the CoCo_Echo (FidoNET): - - - - - - - - - - - - - CUT HERE - - - - - - - - - - - - - - - From tvdogelphi.com Date: Sun, 16 APR 95 23:44:13 -0500 From: "Jeffrey L. Hayes" Newsgroups: alt.bbs.internet Subject: Tandy BBS questionnaire (please reply) Do you run a BBS that supports Tandy computers, any model? If so, please answer this questionnaire. I'm compiling a list of BBS's that support Tandy. The list is to be published periodically on comp.sys.tandy and uploaded to FTP sites. Download this questionnaire and fill it out with your word processor, then email it to me at: tvdogelphi.com Please forward this questionnaire to any sysop you may know of who may wish to reply. Replies can also be sent paper mail to: Jeffrey L. Hayes 130 Forest Ave., Lot 1 Orono, Me. 04473 USA Partially filled-out questionnaires are fine :-). ---------------------------------------------------------------------- TANDY BBS QUESTIONNAIRE 1. What is the name of your BBS? 2. In what city/town is the BBS located (include state, province, or country)? 3. What is the phone number (international format if outside the U.S. and Canada)? 4. Who is the sysop? 5. What baud rates does the BBS support (e.g., 300-14400)? 6. What BBS software do you use (e.g., Wildcat, etc.)? 7. How much does the BBS charge? This includes upload/download ratios, if any. Say as much as you want to here :-). 8. What models of Tandy computers does the BBS support? Check those that apply. [ ] Model I/II/III/IV/4D [ ] Model 100/102/200 [ ] Model 16/6000 or other Tandy Xenix [ ] Color Computer, any model [ ] Tandy 2000 [ ] 1000-series [ ] 2500-series [ ] 3000-series [ ] 4000-series [ ] PC-compatible Tandy laptops/portables [ ] Non-PC-compatible Tandy laptops/portables [ ] Other/additional info. Include any information you want about what kind of Tandy support you offer. 9. Is there anything you would like to add, some kind of advertising blurb for your BBS, perhaps :-)? (Up to 20 lines, 70 spaces per line.) Thanks for your reply. - - - - - - - - - - - - - - CUT HERE - - - - - - - - - - - - - - end of crosspost.... #: 20941 S9/Utilities 04-May-95 13:30:02 Sb: network tar ? Fm: Marc Tritschler 100344,2706 To: all I am an OS9/68K user running 4 VME based development systems (with four separate hard drives). The systems are networked using OS9/INTERNET but I dont have NFS. Hence when I backup my drives I have to manually connect the tape streamer onto the SCSI port of each system in turn and use tar locally. Is there any tar type utilities out there that I can use to allow me to leave the tape streamer on one system only and backup across the network ? From my (very limited) understanding of UNIX utilities I think I am looking for something similar to tar running locally and a utility called rmt to run on the machine with the streamer. Thanks in advance for any responses, Marc. #: 20952 S10/OS9/6809 (CoCo) 12-May-95 16:57:04 Sb: #shellplus2.2a Fm: Rick LaVallie 76653,2244 To: Curtis Boyle/ 71310,1271 Curtis , Hi My name is Rick LaVallie I've just downloaded shell2.2a and installed it .the first thing i've noticed is that the wildcards are defaulted off. Now while reading through some old messages concerning this i see where you wrote to Ian Hodgson that the wildcarding offsets are at $4FA and should be $27 default off $26 default on. now when I ded shell2.2a at offset $4Fa I find that it is set at $C9 now I am wondering if this right if I have a differend version or what? Please check this address again for me thanks A new old COCO user -Rick LaVallie- There is 1 Reply. #: 20956 S10/OS9/6809 (CoCo) 13-May-95 17:38:23 Sb: #20952-#shellplus2.2a Fm: David Breeding 72330,2051 To: Rick LaVallie 76653,2244 (X) > Curtis , Hi My name is Rick LaVallie I've just downloaded shell2.2a and > installed it .the first thing i've noticed is that the wildcards are > defaulted off. Rick, Just a note here. Of course, I suppose it's just a matter of "to each his own" but I never did turn wildcarding on. You can turn it on at any time by preceding the command line with a colon (or off this way if you have it defaulted to on). Many utilities and apps use wildcarding in their parameters, and if you use them with the shell also recognizing it and you can get some problems; it seems more this way than if you don't have it. When you try wildcarding and forget the colon and about all that happens is that you get error 215, or the like and you can go back and do it right. If you have wildcarding on, something wrong might happen and you might not notice it. You can do whatever you like, but you might try it defaulted off and see how you like it. -- David Breeding -- CompuServe : 72330,2051 Delphi : DBREEDING *** Composed with InfoXpress/OSK Vr. 1.02 & VED Vr. 2.4.0 *** There is 1 Reply. #: 20958 S10/OS9/6809 (CoCo) 15-May-95 07:57:50 Sb: #20956-#shellplus2.2a Fm: Rick LaVallie 76653,2244 To: David Breeding 72330,2051 (X) Dave. Thanks for the tip I'll give it a try for a while and see if I like it. I've been useing shellplus2.1 for a long time with wildcards on and have not had any problems but i'll give it a try with them off. May be you know why when I EXIT from TSWORD the menu fmt for TSEDIT my vdg screen locks up this has only started since I have installed Shellplus2.2a.None of the windows lock up and as soon as I kill the process the VDG screen comes back to life. Thanks in advance for any advice you can give. Rick LaVallie There is 1 Reply. #: 20962 S10/OS9/6809 (CoCo) 17-May-95 19:13:25 Sb: #20958-shellplus2.2a Fm: David Breeding 72330,2051 To: Rick LaVallie 76653,2244 (X) > Dave. Thanks for the tip I'll give it a try for a while and see if I like > it. I've been useing shellplus2.1 for a long time with wildcards on and > have not had any problems but i'll give it a try with them off. I just never went to the trouble of turning it on, and just thought it would be easier this way. Either way should be good. Of course, if you are used to it turned on, it might be hard to get used to it the other way. While I was using the Coco, I _normally_ could remember to hit the colon when needed, but now, after being used to wildcarding all the time in OSK, when I go back to the coco, I'm always forgetting.. > May be you know why when I EXIT from TSWORD the menu fmt for TSEDIT my > vdg screen locks up this has only started since I have installed > Shellplus2.2a.None of the windows lock up and as soon as I kill the > process the VDG screen comes back to life. It didn't do this with shell+2.1? My guess would be that it is in the way it is blocking the signals.. If you call this window from another window, try going back to that parent window and doing either "dir" or "procs". If this unlocks it, this would be the cause, IMO.. I'm not very well versed on how the system works at that level, but I would give this a try. I know that often, when a process died, often you would see "DEAD" in the process name for procs (or proc, maybe), till you did this.. It seemed that only certain commands would erase this, and the two above would, if I remember correctly. -- David Breeding -- CompuServe : 72330,2051 Delphi : DBREEDING *** Composed with InfoXpress/OSK Vr. 1.02 & VED Vr. 2.4.0 *** #: 20959 S12/OS9/68000 (OSK) 15-May-95 22:11:22 Sb: #20897-Install Intrpt Handler Fm: Jost Eberbach 73502,2041 To: Taji S. Abraham 76760,2232 Taji, did the code I sent you abouy the ISRs work? Jost #: 20932 S12/OS9/68000 (OSK) 26-Apr-95 20:24:02 Sb: #20921-#Help with VG Machines! Fm: Jost Eberbach 73502,2041 To: Jesus Lazaro 100255,504 (X) Jesus, >> These machines use OS-9 and we need to know how it works<< Seems you hace real problem here. Have you considered hiring a consultant who is knowledgable about OS-9? Actually, the people who you bought those machines from should be able to help you out... Joey There is 1 Reply. #: 20938 S12/OS9/68000 (OSK) 28-Apr-95 16:40:25 Sb: #20932-Help with VG Machines! Fm: Jesus Lazaro 100255,504 To: Jost Eberbach 73502,2041 (X) Jost, Thanks a million for your help, but... I don't know any expert in OS-9 here in Spain. Do you know anoybody? Thanks again. Jesus Lazaro Madrid, Spain #: 20943 S12/OS9/68000 (OSK) 05-May-95 17:59:12 Sb: Help with VG Machines! Fm: Jesus Lazaro 100255,504 To: Jost Eberbach 73502,2041 (X) Jost, Thanks again for your help. Our Tech Department is working very hard and they have some solutions, so we hope to do all work in a couple of weeks. Thanks a million. Jesus #: 20923 S12/OS9/68000 (OSK) 25-Apr-95 06:16:54 Sb: #Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: all I appear to have a problem with OS-9 alarms 'disappearing'. After a successfull call to alm_set() I get no signal returned when the alarm expires. Also, I have a utility that shows what the system is up to, including a list of active alarms. The alarm just set doesn't show up, as if it was never successfully set. The function call, however, returns a valid alarm id. Has anyone else experienced a problem like this? The same code can work for 10's of iterations, then fail. I'm using v2.4.4 of OS-9 and Ultra 'C' v1.1. I've tried old and new-style function calls, both with the same effect. Like all itermittent problems, this is proving a real pain. (It took half a day to get the system to misbehave while I was monitoring it.) Any help VERY welcome. There is 1 Reply. #: 20924 S12/OS9/68000 (OSK) 25-Apr-95 07:47:39 Sb: #20923-#Alarms 'disappearing' Fm: Joachim Terasa 100421,2472 To: Ian J Shearer 100410,2733 (X) >> I appear to have a problem with OS-9 alarms 'disappearing'. << What you are talking about seems to be a similar problem I had 2 years ago also with OS-9 2.4. In my case it was not an alarm problem rather than a problem of lost signals. My application had to wait for characters from several pipes. So each pipe was set up with a signal. In some very rare cases it happened that two signals were generated within a very short time. That means, the second signal appeared right in the moment when the process was leaving the sleep(0). The mechanism as shown in the examples that store the signal number in a global variable inside the signal handler and the evaluates it inside the process was not usable any more. Do you see something you can use? Jochen -Joachim Terasa There are 2 Replies. #: 20926 S12/OS9/68000 (OSK) 25-Apr-95 14:50:59 Sb: #20924-#Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: Joachim Terasa 100421,2472 (X) Thanks for the advice, but I think I'm seeing something different. Two reasons; first, the utility that shows what is happening in the system doesn't recognise that the alarm is ever being set. Second, I've tried putting a printf() in the signal handler (not good practice, but really useful for debugging) and this doesn't see the signal ever arriving. I'm told that it is impossible for OS-9 to 'drop' a signal. If one is sent it is ALWAYS seen by the signal handler. The only way to loose the information is by using the normal technique of setting a global variable, then finding that you've over-written it before it is processed. This is the problem you saw. If my reasoning is correct the signal never gets to the process because the alarm isn't correctly set. Does this make sense?? Thanks anyway; any more advice gratefully received. There is 1 Reply. #: 20927 S12/OS9/68000 (OSK) 25-Apr-95 20:15:21 Sb: #20926-#Alarms 'disappearing' Fm: Kevin Darling 76703,4227 To: Ian J Shearer 100410,2733 (X) Hi Ian, How about posting a sample of the code you're using to set the alarm and to process it? Perhaps someone here will notice something. best - kev There is 1 Reply. #: 20931 S12/OS9/68000 (OSK) 26-Apr-95 13:40:00 Sb: #20927-Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: Kevin Darling 76703,4227 (X) Kev The code is so trivial there's not a lot to say about it; it's shown below. As I said initially, it DOES work for many iterations before a failure occurs. Two functions are shown. The first sets the timer going, the second is the signal handler. The function setTimer() is just a wrapper for the OS-9 function call. The handler does the usual of setting a global variable and masking out further signals until this one is dealt with. (The printf statement was included when I first saw the problem.) If you can see anything wrong with this I'd love to know; it appears to be so simple it's hard to see where it can fail. Ian. /*****************************************************************************/ int setTimer(int i_Event, int i_Time) /***************************************************************************** * Sets an OS-9 alarm going to generate a signal after the specified time. * * Inputs: i_Event signal/event to be generated * i_Time time interval (in 1/10 of a sec.) before * signal generated * * Returns: -1 if an error occurs * 0 else *****************************************************************************/ { register int i = 0; int i_err, i_rval, i_AlarmID; /* * Set an OS-9 alarm for the desired signal/time */ if ((i_err = _os_alarm_set(&i_AlarmID, i_Event, ALM_TIME(i_Time))) != 0) { err_report(i_err,__LINE__,__FILE__); i_rval = -1; } return i_rval; } /*****************************************************************************/ void handler(int i_signo) /***************************************************************************** *OS-9 signal handler. If the received signal is SIGQUIT, set flag to *abort process. * * Inputs: i_signo OS-9 signal received * * Returns: --- * *****************************************************************************/ { switch (i_signo) { case SIGQUIT: i_ProzessEndeFlag = TRUE; break; case SIG_SIMTRACE: i_Trace = TRUE; break; default: printf("<<%d>>",i_signo); fflush(NULL); e_Event = (BiteEvent)i_signo; sigmask(TRUE); break; } } #: 20928 S12/OS9/68000 (OSK) 26-Apr-95 03:00:48 Sb: #20924-#Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: Joachim Terasa 100421,2472 (X) Another thought about the situation you describe. Normally, you would call sigmask(1) from within the signal handler routine to make sure that signals are masked out until the next time you reach the sleep(0) statement. This SHOULD make sure that if two signals arrive in quick succession the second one doesn't get lost. Were you doing this? There is 1 Reply. #: 20929 S12/OS9/68000 (OSK) 26-Apr-95 04:17:50 Sb: #20928-#Alarms 'disappearing' Fm: Joachim Terasa 100421,2472 To: Ian J Shearer 100410,2733 (X) I solved the problem exactly this way. But don't be sure that the sleep(0) is the only sleep in your software. There are several system calls, especially i/o-services including sleeps implicitly. So it's really not trivial to protect your sleep/signal-mechanisms from side effects. And it is getting chaotic if you are not only using sleep/signal but additionally events in one process! -Joachim Terasa There is 1 Reply. #: 20930 S12/OS9/68000 (OSK) 26-Apr-95 13:22:30 Sb: #20929-#Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: Joachim Terasa 100421,2472 (X) I am using events as well. What has surprised me, is that the problem still occurs if I reorganise my function calls to make sure that setting an alarm is the last thing I do before sleeping. If it wasn't for this I'd believe that ev_wait() or some such was waking up without the signal being fully processed. Ian There is 1 Reply. #: 20933 S12/OS9/68000 (OSK) 27-Apr-95 02:39:31 Sb: #20930-#Alarms 'disappearing' Fm: Joachim Terasa 100421,2472 To: Ian J Shearer 100410,2733 (X) >> If it wasn't for this I'd believe that ev_wait() or some such was waking up without the signal being fully processed. << Hanging on an _ev_wait() the function will awake with a signal, the signal will be completely processed. You would have to check if the event condition is really true after the _ev_wait. Hanging on a sleep() an event will not affect your program. Jochen -Joachim Terasa There is 1 Reply. #: 20968 S12/OS9/68000 (OSK) 24-May-95 13:22:10 Sb: #20933-#Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: Joachim Terasa 100421,2472 (X) If you're interested, I found the cause of my problem. If you delete an alarm at the same time as it expires, it is possible to accidentally delete a completely different alarm from another process. It goes something like this; Task A starts to delete an alarm The alarm being deleted expires A task/process switch occurs Task B creates a new alarm, which uses the old 'slot' of the task A alarm Task A completes deleting the alarm, but gets the new alarm from task B by mistake This is something to do with OS-9 marking an alarm 'slot' as free before it truely is. Version 3.0 gives a partial solution, since you can prevent a task from deleting alarms with other owner IDs; if you spawned all the tasks from one parent, as I did, this won't help. The workaround is to use periodic (repeating) alarms, then kill them in response to the first signal they generate. In other words, the ONLY safe way to use alarms is to avoid single-shot alarms like the plague. Disappointing but true. There are 2 Replies. #: 20970 S12/OS9/68000 (OSK) 26-May-95 02:28:54 Sb: #20968-Alarms 'disappearing' Fm: Joachim Terasa 100421,2472 To: Ian J Shearer 100410,2733 (X) >> In other words, the ONLY safe way to use alarms is to avoid single-shot alarms like the plague. << Thanks very much for this info. I'll keep it in mind! -Joachim Terasa #: 20976 S12/OS9/68000 (OSK) 31-May-95 17:24:34 Sb: #20968-Alarms 'disappearing' Fm: Joerg Hofstetter 100112,2561 To: Ian J Shearer 100410,2733 I made the same experience with OS-9 alarms: Single-Shot alarms are sometimes disappearing. I always use alm_cycle(). It's great that you found an explanation. Thank you. I hope Microware is also reading ..... #: 20944 S12/OS9/68000 (OSK) 08-May-95 15:59:45 Sb: #HP Jet Direct Net-Interf Fm: Alptekin Aydogan 100527,324 To: all Who has experience with HP Jet Direct EX Network Interface for printers ? I'm using OS9 Ver. 2.4 with TCP-IP + NFS and would like to connect my OS9-developing system to a HP Deskjet-Printer which has this interface. There is 1 Reply. #: 20955 S12/OS9/68000 (OSK) 13-May-95 13:54:20 Sb: #20944-HP Jet Direct Net-Interf Fm: Joachim Terasa 100421,2472 To: Alptekin Aydogan 100527,324 (X) Hi, >> Who has experience with HP Jet Direct EX Network Interface for printers ? << Would also be helpfully for me! If you've found a solution - let me know. Because it is a thin based on TCP/IP-sockets it may be portable. I will also keep my eyes open! Jochen -Joachim Terasa #: 20947 S12/OS9/68000 (OSK) 11-May-95 18:24:45 Sb: #How to link I2C to OS-9 Fm: Martin Kreutzer 100020,2550 To: all I have to port an OS-9 to a target system that has some devices on a I2C-Bus (driven by a PCF8584). Now I wonder what`s the "OS-9-like" way to link I2C to the System. I2C has commands like "write 2 bytes, then read 5 bytes answer" I see three ways to handle this: 1) leave the I/O System alone and make a system-state module to handle I2C. 2) use the I/O System. Delay write requests until I get the corresponding read request. I see some problems to keep the whole procedure atomic. Furthermore I think I have to write a new filemanager, because I need the sizes of the I/O requests in the device drivers. 3) use the I/O System but handle the requests in SetStat calls. If I handle pure reads and writes with SetStats too, I can connect the device driver to a standard Filemanager (e.g. SCF). Right now I prefer Version 3 but I`m open to suggestions. Please give me your opinion.. Martin Kreutzer There are 3 Replies. #: 20948 S12/OS9/68000 (OSK) 11-May-95 19:58:35 Sb: #20947-How to link I2C to OS-9 Fm: Jost Eberbach 73502,2041 To: Martin Kreutzer 100020,2550 (X) Martin, If I were to do that, I would definitely prefer version 1, >>leave the I/O System alone and make a system-state module to handle I2C.<< If you do it yourself, you can at least be sure, that OS-9 doesn't mess with your data! It also seems to be the simplest methods. Of course, the other versions are more challenging... I have broken my head over OS-9 too much I guess, I really try not to use it at all if possible. I do all my hardware related stuff in system state modules and/or ISRs, I don't bother to try and write an OS-9 driver etc. Just my opinion... Jost #: 20949 S12/OS9/68000 (OSK) 12-May-95 01:38:19 Sb: #20947-How to link I2C to OS-9 Fm: Joachim Terasa 100421,2472 To: Martin Kreutzer 100020,2550 (X) >> Please give me your opinion.. << I also prefer version 3, especially if you don't have to develop a device driver from the scratch, but just have to modify something existing. If you are not experienced in writing OS-9 device drivers version 1 would be the better way. Jochen -Joachim Terasa #: 20953 S12/OS9/68000 (OSK) 12-May-95 21:40:36 Sb: #20947-How to link I2C to OS-9 Fm: Bob van der Poel 76510,2203 To: Martin Kreutzer 100020,2550 (X) You might want to contact Kevin Darling on this forum...I believe he wrote (or started to write???) a driver for the i2c chip on the mm/1. Kevin? #: 20964 S12/OS9/68000 (OSK) 18-May-95 13:23:34 Sb: #buffers Fm: LARRY OLSON 72227,3467 To: Kevin Darling 76703,4227 (X) Kevin, I have a question about Kwindows get/put buffers, that I hope you could help me with. It concerns allocating memory for the get/put buffers. How do I allocate memory for an entire GROUP. I know of the 1B 29 GRP BUF HCT LCT , function, but I have about 50 separate object buffers under 1 group(P_ID) number. These buffers vary in size from 3 x 3 to 15 x 10. Do I need to define each one, or is there a way to allocate memory for the entire GROUP ? Does the function work like KILBUF, where using 0 for the buffer number applies to all buffers in the GROUP ? I thought of splitting some of the buffers off into another GROUP, like (P_ID + 1) but wouldn't you then run the risk of hitting the GROUP number of another process I hope you can set me straight, Larry There is 1 Reply. #: 20971 S12/OS9/68000 (OSK) 28-May-95 13:17:42 Sb: #20964-#buffers Fm: Kevin Darling 76703,4227 To: LARRY OLSON 72227,3467 (X) Hi Larry, Unfortunately, there isn't a way to allocate memory for a group of buffers ahead of time. The internal system has to know where each "object" will begin so it can make better decisions about storing it. If you run out of buffers, some people use P_ID+128, although that may not be safe on everyone's system. best - kev There is 1 Reply. #: 20975 S12/OS9/68000 (OSK) 31-May-95 12:47:48 Sb: #20971-buffers Fm: LARRY OLSON 72227,3467 To: Kevin Darling 76703,4227 (X) > Unfortunately, there isn't a way to allocate memory for a group of > buffers ahead of time. The internal system has to know where each > "object" will begin so it can make better decisions about storing it. > > If you run out of buffers, some people use P_ID+128, although that may not > be safe on everyone's system. Kevin, Thanks, I'll keep that P_id + 128 in mind. What brought the question up was as usual, another dumb mistake of mine. I kept getting the error message - can't allocate memory -, when I was trying to run this program and I thought the problem was in allocating memory for the GET/PUT buffers. I had forgotten that I had changed another part of the program, that was loading a sound file. Well now the sound file wasn't where the program thought it was and that was where the program kicked out the -can't allocate memory- error. Ok, I'll get back to playing with these GET/PUT buffers again, I havn't got it figured out completely yet but the GET appears to be sensitive about where the object is on the screen, because some objects when they are later put to the screen, the rightmost pixels of the object are missing. I thought I had it figured out by drawing the objects so that they fall on the 8 bit boundries, and for some of the objects this worked, but not for all. Thanks again Kevin larry #: 20973 S12/OS9/68000 (OSK) 30-May-95 13:22:57 Sb: #Alarms 'disappearing' Fm: Paul N. Leonard 72477,774 To: 100421,2472 (X) I just now saw your messages about disapearing alarms. It is true that you can unintentionally delete an alarm but it is definitely avoidable. Rather than use the sequence sited in your messages use the sequence shown below and you won't have any problems. You must be consistent, however, because if you do it incorrectly in one place, you are back where you started. For each alarm that is created set a unique flag to indicate it is active. When the alarm handler receives that alarm it clears the proper flag to indicate the alarm has fired. When it comes time to delete an alarm set the signal mask to prevent alarms from firing. Check the status of the flag. If it is still set, then it has not fired and you can safely delete the alarm and clear the flag. If it is clear, then there is no need to clear the alarm or flag because it has already fired. Clear the signal mask so signals can be received again. We are using this scheme on our product which can have up to 24 tasks running at one time of which 10 are creating and deleting alarms a great deal of the time. We had the same problem until we came up with this solution and it hasn't been a problem since. Hope this helps. Paul N. Leonard There is 1 Reply. #: 20974 S12/OS9/68000 (OSK) 30-May-95 14:10:52 Sb: #20973-Alarms 'disappearing' Fm: Ian J Shearer 100410,2733 To: Paul N. Leonard 72477,774 (X) Paul, Thanks for the tip; I see how it works. What you're doing is basically compensating for a weakness in OS-9; have you any idea why there is no internal mechanism for preventing the problem? Microware are obviously aware of the difficulty. Unfortunately the 'fix' in Version 3.0 is of very limited value. Ian Press !>